JPH11120040A - Method for predicting effect of parallel processing procedure, and recording medium therefor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To automatically predict and output the effect of a parallel processing procedure by a computer. SOLUTION: To predict and diagnose the effect when the parallel processing executable processing procedure consisting of steps is processed in parallel by the computer, a parallel processing effect predicting device 1, collects SMF (system management function) records when the processing procedure are processed in series from an SMF data set 2 stored on a magnetic disk, analyzes the SMF records to calculate the serial processing elapsed time of the serial processing of the processing procedure, the predicts and calculates the deletion time at the time of the parallel processing of the processing procedure, and substrates the deletion time from the serial processing elapsed time, thus predicting and calculating a parallel processing elapsed time.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for predicting the effect of a parallel processing procedure by an electronic computer and a recording medium for the same, and more particularly, to a method for executing a parallel processing procedure (job step) comprising a plurality of steps. The present invention relates to a method for estimating the effect of a parallel processing procedure for estimating and outputting the elapsed execution time of a processing procedure processed by a computer when a system storage area (pipe data set) is applied to take over the data, and a recording medium therefor.

[0002]

2. Description of the Related Art Conventionally, a JC that describes a procedure instruction has been described.
In a processing procedure composed of a plurality of steps described in L (job control statement) (hereinafter simply referred to as a job step), each step is executed (serial processing) in the order described.
Generally, in a computer that can process a plurality of instructions in parallel, a plurality of job steps are processed in parallel to improve efficiency. In particular, when a subsequent job step inputs and processes data of a data set output by a preceding job step, a virtual file, i.e., pipe data, is stored in system storage without using a direct access storage device or a magnetic tape device. The method of securing a set is adopted. In this specification, a batch process to which the pipe data set is applied is called an Excel batch, and a batch process to which the pipe data set is not applied is called a normal batch.

Japanese Patent Laying-Open No. 9-6628 discloses an apparatus and method for selecting a parallel processing procedure proposed by the present applicant. This apparatus is characterized in that it comprises a parallelization checking means for checking a method of using a data set of a continuous processing procedure and selecting a processing procedure which can be executed in parallel, particularly by judging a pipe data set. Next, the pipe data set will be described below.

FIG. 12 is a diagram showing data transfer between parallel job steps using a pipe data set.
FIG. 12 shows a system storage file (pipe data set) used for transferring data between job steps to be executed in parallel, a job step 1 for writing data via the pipe data set, and a data reading side. Job steps 2 and 3 are shown.

The data B1, B2,..., Bn created and output by the writing-side job step 1 are immediately passed to the reading-side job steps 2 and 3 via the pipe data set. After the data is passed to the job steps 2 and 3 on the reading side, the data in the pipe data set is deleted. Therefore, no matter how much data is taken over, a small area (64 K to several 100 K bytes) in the system storage is used. ), The system storage can be effectively used.

At the time of opening the first pipe data set in job steps 1 to 3 (if job step 2 is first opened), a system storage file for the pipe data set is acquired on the system storage. In the case other than the first opening, the access environment to the system file for the pipe data set is prepared.

If the reading side (eg, job step 2) opens first and starts reading, but the writing side (job step 1) is not operating or has not started writing, the reading process is performed. Now wait for the writer to write the data. On the writing side, B1, B2, ...,
Data is written to the pipe data set in the order of Bn, Bn + 1,... Data is temporarily stored on a system storage file.

The reading side (job steps 2 and 3) becomes readable at the same time data is written.
Data is read in the order of 2, B3,. The same data is passed to all reading job steps. When the same data is read in all reading-side job steps, the area on the system storage file storing the data is released. The area on the system storage file is used cyclically.

When writing data, if the pipe data set is full of accumulated data, the writing job step waits until the reading side reads the data and the pipe data set is free. If there is no data on the pipe data set when reading data, the reading job step waits until the writing job step writes data. Thus, the writing side (job step 1), the reading side (job steps 2, 3)
Data can be transferred while being executed asynchronously in parallel.

[0010]

When the batch processing using the pipe data set as described above, that is, the Excel batch is applied, how much the processing efficiency by the computer is improved, that is, the desired job step is changed to the Excel batch When the parallel processing is performed, the computer user can reduce the processing time of the computer compared with the case of the serial processing and predict the effect of the parallel processing procedure of how much processing time is required. Manual calculations must be performed based on SMF (system management function) records related to job steps. The SMF is a function that enables a computer system to perform a job schedule unique to the computer system, and collects and records information on various events such as a start time and an end time when a computer executes a predetermined job. There is a problem that the user has to spend a lot of time and effort for this manual calculation.

SUMMARY OF THE INVENTION It is therefore an object of the present invention to solve the above problems and to provide a parallel processing procedure prediction method for automatically predicting and outputting the effect of a parallel processing procedure by a computer, and a recording medium therefor.

[0012]

FIG. 1 is a basic configuration diagram of the present invention. A method for estimating the effect of a parallel processing procedure according to the present invention that solves the above problem is a method for estimating the effect of a parallel processing procedure in which a parallel processing executable processing procedure including a plurality of steps is processed by a computer in parallel. The prediction device 1 collects, for example, an SMF (system management function) record from the SMF data set 2 stored on a magnetic disk when the processing procedure is serially processed, analyzes the SMF record, and serializes the processing procedure. Calculate the serial processing elapsed time at the time of processing, predict and calculate the reduction time when the processing procedure is performed in parallel from the SMF record, subtract the reduction time from the serial processing elapsed time, and perform parallel processing by the parallel processing. The method includes the steps of predicting and calculating the processing elapsed time, and predicts the effect of executing the parallel processing procedure in parallel. Characterized in that the diagnosis.

The method for predicting the effect of a parallel processing procedure according to the present invention may further include, for example, editing the elapsed time of the parallel processing and the reduction time as a result of the prediction diagnosis of the effect of parallel processing of the parallel processing procedure, and predicting the effect on a printer or a CRT. Outputting a list 5. According to the above configuration, the effect of the parallel processing procedure by the computer can be automatically predicted and output.

A method for estimating the effect of a parallel processing procedure according to the present invention which solves the above-mentioned problem has a plurality of processing procedures which collect a plurality of processing procedures capable of executing parallel processing consisting of a plurality of steps. Is a parallel processing procedure for performing, by a computer, a parallel processing procedure group consisting of a first parallel processing procedure group for taking over data in a pipe data set and a second parallel processing procedure group for not taking over data in a pipe data set. In the group effect prediction method, the SMF when the parallel processing procedure group is serially processed
A record is collected, a serial processing elapsed time when the parallel processing procedure group is serially processed is calculated from the SMF record, and a first elapsed time when the first parallel processing procedure group is parallel processed based on the SMF record. Is calculated by summing the respective predicted elapsed times of the respective processing procedures forming the first parallel processing procedure group, and the second elapsed time when the second parallel processing procedure group is processed in parallel is referred to as the second parallel processing procedure group. Determined from the maximum value of the estimated elapsed time of each of the processing procedures to be performed,
Calculating the parallel processing elapsed time of the parallel processing procedure group from the maximum value of the obtained first elapsed time and the second elapsed time;
Each step of subtracting the parallel processing elapsed time from the serial processing elapsed time to predict and calculate a reduced time reduced by the parallel processing, and predicting and diagnosing an effect of performing the parallel processing procedure group in parallel. And

The method for predicting the effect of a parallel processing procedure according to the present invention may also include, for example, editing the parallel processing elapsed time and the reduction time as a result of predicting and diagnosing the effect of performing the parallel processing procedure group on a printer or CRT. Output step. The method for predicting the effect of a parallel processing procedure according to the present invention also includes a processing time for serially processing a complex processing procedure having a plurality of the parallel processing procedures by a computer, and a parallel processing of each parallel processing procedure group forming the complex processing procedure. Calculate the total elapsed time.

According to the method for predicting the effect of a parallel processing procedure according to the present invention which solves the above problem, a processing procedure capable of executing parallel processing comprising a plurality of steps comprises a first step comprising a plurality of steps for taking over data using a pipe data set. In a method for predicting the effect of a parallel processing procedure in which a first group of steps is processed in parallel by a computer when a group of steps and a second group of steps that do not take over data in the pipe data set are performed, The SMF record obtained when the serial processing is performed is collected, and the SMF record is collected.
The first when the first step group is processed in parallel from the record
The method includes the steps of predicting and calculating the reduction time of the step group, and predicting and diagnosing the effect of parallel processing of the first step group.

The method for predicting the effect of the parallel processing procedure according to the present invention also edits, for example, the parallel processing elapsed time and the reduced time as a result of the prediction diagnosis of the effect of performing the first step group in parallel, and outputs the result to a printer or CRT. Performing the steps of: In the method for predicting the effect of a parallel processing procedure according to the present invention, when the processing procedure comprises a plurality of first step groups, the reduction time of the processing procedure is calculated by summing up the respective reduction times of the first calculated step groups. The time is predicted and calculated, the serial processing elapsed time when the processing procedure is serially processed from the SMF record is calculated, and the parallel processing when the processing procedure is processed in parallel by subtracting the reduction time from the serial processing elapsed time is calculated. Each step of predicting and calculating the processing elapsed time is provided.

In the method for predicting the effect of a parallel processing procedure according to the present invention, in addition to the analysis of the collected SMF record, a JC describing instructions of each step of the processing procedure is described.
The method further includes a step of analyzing the L (job control statement), for example, the JCL library 3 stored on the magnetic disk, and determining a group of steps for taking over the data using the pipe data set.

When a procedure library 4 stored on a magnetic disk and called from the JCL, for example, is input in addition to the JCL library 3, the accuracy of effect prediction is improved. A computer-readable recording medium according to the present invention that solves the above-mentioned problems is a recording medium for causing a computer to execute a method for predicting the effect of a parallel processing procedure in which a computer performs parallel processing executable processing procedures including a plurality of steps. In the medium, an SMF record obtained when the processing procedure is serially processed is collected by the computer, and the SMF record is collected.
Calculating the serial processing elapsed time when the processing procedure is serially processed from the record, predicting and calculating the reduction time when performing the processing procedure in parallel from the SMF record;
A program for predicting and diagnosing the effect of parallel processing of the parallel processing procedure for executing each step of predicting and calculating the parallel processing elapsed time by the parallel processing by subtracting the reduction time from the serial processing elapsed time is recorded. It is characterized by.

Further, the recording medium of the present invention stores, in a storage means of a computer, a prediction diagnosis result of an effect of performing the parallel processing procedure in parallel, for example, the parallel processing elapsed time and the reduced time, and edits the parallel processing procedure in a printer or CRT. This is a recording medium on which a program for executing an output step is recorded.

[0021]

Embodiments of the present invention will be described below with reference to the accompanying drawings. FIG. 2 is a flowchart of a method for predicting the effect of a parallel processing procedure according to the present invention.
In this flowchart, as an example, a processing procedure that can execute parallel processing including a plurality of steps, that is, processing of effect prediction when job steps are processed in parallel is shown. Hereinafter, the flowchart of FIG. 2 will be described with reference to the basic configuration of the present invention shown in FIG.

First, in step S 1, the parallel processing effect prediction device 1 sets the SMF input from the SMF data set 2
Analyze IN (SMF data). This SMF data analysis will be described later. In step S2, the JCL
It is determined whether or not there is a JCLLIB (input JCL) from the library 3, and if the result of the determination is YES, step S
The process proceeds to step S3, and if NO, the process proceeds to step S4. Also, J
If a PROCLIB (procedure) 4 called from the JCL is input in addition to the input JCL from the CL library 3, the applicability inspection of the pipe data set (application batch application diagnosis) can be performed with high accuracy. The accuracy of the effect prediction is improved. Step S3
Then, the parallel processing effect prediction device 1 analyzes the input JCL, and generates a JCL in step S4.

In step S5, the parallel processing effect prediction device 1 analyzes the application diagnosis of the Excel batch. In step S6, it is determined from the application diagnosis of the Excel batch in step S5 whether or not the Excel batch is applicable. If the determination result is YES, the process proceeds to step S7, and if NO, the present routine ends. In step S7, the effects of the Excel batch are predicted from the SMF data analysis in step S1 and the Excel batch application diagnosis in step S5. In step S8, a diagnosis list of the effect prediction of the Excel batch, EFCTLIST (effect prediction list) 5 shown in FIG. 1 is output. The effect prediction list includes a step group (data set) unit, a parallel processing procedure (job step) unit, a job step group (processing procedure group) unit, a parallel job step group (parallel processing procedure group) unit, and a complex processing procedure ( Job) unit.

Here, the recording medium of the present invention will be briefly described. The parallel processing effect prediction device 1 shown in FIG.
A reading device (not shown) for a recording medium such as a ROM or a floppy disk, and various programs related to the method according to the present invention recorded on a recording medium such as a CD-ROM or a floppy disk by a predetermined operation from the reading device. The data is read and loaded into an auxiliary storage device (not shown). Of course, the operator may write the program directly into the main storage device of the parallel processing effect prediction device 1, for example, the RAM, and then load the program into the auxiliary storage device. After that, the program stored in the auxiliary storage device is
The data is loaded into the main storage as required by the processing of the CPU (not shown) of the parallel processing effect prediction device 1.

Next, SMF data analysis, that is, SMF
A method for predicting the effect of the parallel processing procedure from the record will be described below. The following data is included in the SMF record when the job steps for which the reduction time and the processing time are predicted by the parallel processing are serially processed. (A) For SMF type 30 records: 1) Record creation date / record creation time (used for calculating execution elapsed time) 2) Start date / start time (used for calculating execution elapsed time) 3) CPU usage time 4 ) Swap-in time 5) Input / output count of job step (b) SMF type 14/15 record 1) Input / output count of DD statement unit 2) Block length (data size: unit byte) Predicted from such SMF record The reduction time and processing time of the job step in parallel processing are calculated as follows.

First, since the swap-in time is the sum of the CPU use time and the input / output processing time, the elapsed time per input / output (I / O) in a job step including a plurality of steps is calculated from the following equation. I do. Average I / O time = (JBS swap-in time-JBS
CPU usage time) / JBS input / output count (Equation 1) In the above equation, JBS indicates a job step.

The estimated reduction time for each DD sentence is calculated from the following equation using the average I / O time calculated by Equation 1 and the number of times of input / output for each DD sentence. Predicted reduction time per DD statement = Average I / O time × Number of input / output times per DD statement ... (Equation 2) A predicted reduction time and a predicted elapsed time per job step are calculated by the following equation.

Predicted reduction time per job step = DD
Total predicted reduction time per statement ... (Equation 3) Estimated elapsed time per job step = Elapsed time-Predicted reduction time ... (Equation 4) Estimated reduction time per parallel job step group diagnosed as applicable to Excel batch The estimated elapsed time is calculated as follows.

-1. Since the elapsed time is not shortened by the parallel processing between the job steps that are taking over the data via the pipe data set in the parallel job step group, the estimated elapsed time of each job step group is calculated by the following equation. . Estimated elapsed time of job step group = sum of estimated elapsed time of each job step group (Equation 5) -2. 100% parallel processing is performed between job steps in the parallel job step group that are not taking over data via the pipe data set.
Among the estimated elapsed times of the job step group calculated by the equation 5, the maximum value is calculated as the estimated elapsed time of the parallel job step group.

The estimated reduction time and the estimated elapsed time for each job are calculated as follows. The estimated elapsed time for each job is calculated by summing the estimated elapsed time for each parallel job step group forming a job. The estimated reduction time for each job is obtained by subtracting the calculated estimated elapsed time from the elapsed time when serial processing obtained from the SMF record is performed.

Finally, a predicted reduction time and a predicted elapsed time in a data set unit are calculated as follows. The predicted reduction time for each data set is calculated as the predicted reduction time for each DD sentence calculated by Expression 2. The estimated elapsed time in the data set unit is obtained by calculating the elapsed time when a plurality of steps constituting the data set are serially processed from the SMF record, and subtracting the estimated elapsed time in the data set unit from the calculated elapsed time. Ask.

Next, the DD sentence used to predict the effect of the Excel batch of the present invention will be briefly described below. a) SMF Data Set DD Sentence: SMFIN DD Sentence This specifies the data set that stores the SMF record of the job whose effect is to be predicted.

B) Input DD statement for JCL library: JC
LLIB DD statement This specifies the data set that stores the JCL of the job whose effect is to be predicted. c) Input DD statement for procedure library: PROCLI
BDD sentence In this way, when inputting a JCL, a data set storing a procedure called from the JCL is specified.

D) DD statement for output of effect prediction list: EFC
TLIB DD statement This specifies a data set for outputting the result of the effect prediction. FIG. 3 is a configuration diagram of the effect prediction list according to the embodiment. The header is output on the first line, and information on the effect prediction diagnosis result is output on the third and subsequent lines. FIG. 4 is a diagram showing an output example of the header row shown in FIG. The upper row shows the column numbers. The first line contains the program name, PDCF TOOL V12L10,
EFFE indicating that this list is a predictive diagnosis list
The date and time when CT-LIST and list output were executed are shown.

FIG. 5 is a diagram showing the output contents of the effect prediction diagnosis result information. This corresponds to the third and subsequent lines of the effect prediction list shown in FIG. The first row from the top shows the column numbers. The second row shows the job name, whether or not to apply Excel, whether or not the effect can be predicted, the start time, the end time, the elapsed time, the CPU usage time, the predicted elapsed time, the predicted reduction time, and the predicted reduction rate of the elapsed time. .

The third row shows the parallel job step group name, the number of parallel job steps, the start time, the end time, the elapsed time, the CPU usage time, the estimated elapsed time, the estimated reduction time, and the estimated reduction rate of the elapsed time. It is. Third
The second row is not output when the SMF data is invalid.
The fourth row shows the job step number, job step name, average I / O (input / output) time, group name, start time, end time, elapsed time, CPU usage time, estimated elapsed time, estimated reduction time, and elapsed time. The predicted time reduction rate is shown. The fourth row is not output when the SMF data is invalid.

The fifth row shows the data set name, whether or not to apply the pipe data set, the result of the block size optimization diagnosis, the job step number of the job step using the data set, the DD name of the DD statement, and the prediction reduction. The predicted reduction rate of time and elapsed time is shown. The fifth stage is SMF
It is not output when the data is invalid. FIG. 6 is a diagram illustrating an output example of the effect prediction diagnosis result information for each job.
(A) is a diagram showing the left half, and (B) is a diagram showing the right half. FIG. 6 shows (a) the job name, (b) the use of Excel (the JCL analysis / interpretation resulted in a JCL error or the SMF data was invalid, and it was decided that no Excel was used), and (c) the effect. Predictability (JCL analysis / interpretation determines JCL error or invalid SMF data determines effect prediction not possible), (d) Job start time, (e) Job end time, (f) elapsed time,
(g) CPU usage time, (h) predicted elapsed time, (i) predicted reduced time, (j) predicted reduction rate of elapsed time, and (k) reason for failure of effect prediction diagnosis. If a JCL analysis / interpretation results in a JCL error or invalid SMF data, the reason is indicated.

FIG. 7 is a diagram showing an output example of the effect prediction diagnosis result information for each parallel job step group. FIG. 7 shows (a) the parallel job step group name, and (b)
Number of parallel job steps, (c) parallel job step start time, (d) parallel job step end time,
(e) Elapsed time, (f) CPU usage time, (g) Estimated elapsed time, (h) Estimated reduction time, and (i) Estimated reduction rate of elapsed time are shown.

FIG. 8 is a diagram showing an output example of the effect prediction diagnosis result information for each job step. FIG. 8A is a diagram showing the left half, and FIG. 8B is a diagram showing the right half. FIG. 8 shows (a) the job step number, (b) the job step name,
(c) Average I / O (input / output) time (unit: seconds) for the entire job step, (d) Parallel job step group name including the job step, (e) Start time of job step, (f) The job step end time, (g) elapsed time, (h) CPU usage time, (i) predicted elapsed time, (j) predicted reduction time, and (k) predicted reduction rate of elapsed time are shown.

FIG. 9 is a diagram showing an output example of effect prediction diagnosis result information in data set units. FIG. 9 shows (a) data set name, (b) whether or not pipe data set is applied, (c) block size optimization diagnosis result, (d) job step number of the job step using the data set, ( e) DD name of DD statement that defines the data set, (f) Predicted reduction time of job step by applying pipe data set of data set and block optimization, and (g) Application of pipe data set of data set and block optimization The estimated reduction rate of the elapsed time of the job step due to the conversion is shown.

FIG. 10 is a diagram showing an output example of the effect prediction list. As described with reference to FIGS. 3 to 5, the header is output on the first line, and the information on the effect prediction diagnosis result is output on the third and subsequent lines. Next, the pipe data set applicability check will be briefly described. FIG. 11 is a diagram showing a specific example of the JCL for parallel job steps. Each command statement of a job step describes a job step number, a DD statement number (written as necessary), and a JCL.
Includes descriptions of ARA GRP (parallel group parameters) and PIPE (pipe data set parameters). FIG.
In the JCL as shown in the figure, pipe data set applicability check is performed. If the data set is defined only by the 1DD statement in the job, if no new data set is created, the same name When a plurality of data sets are defined by DD statements, it is determined that the pipe data set is not applicable. As described above, the pipe data set applicability test can be performed with high accuracy by the JCL, and therefore, a more accurate prediction diagnosis can be performed than when only the SMF record is used to predict the elapsed time or reduced time due to parallel processing of job steps.

In the embodiment of the present invention described above, the list output by the printer has been described as an example. However, the present invention is not limited to this, and the output may be displayed on a CRT.

[0043]

As described above, according to the present invention,
It is possible to provide a method of predicting a parallel processing procedure for automatically predicting and outputting the effect of a parallel processing procedure by a computer, and a recording medium therefor. As a result, a user of the computer can significantly increase the time and effort for the prediction. Can be reduced.

[Brief description of the drawings]

FIG. 1 is a basic configuration diagram of the present invention.

FIG. 2 is a flowchart of an effect prediction method of a parallel processing procedure according to the present invention.

FIG. 3 is a configuration diagram of an effect prediction list according to the embodiment.

FIG. 4 is a diagram showing an output example of a header row shown in FIG. 3;

FIG. 5 is a diagram showing output contents of effect prediction diagnosis result information.

FIG. 6 is a diagram showing an output example of effect prediction diagnosis result information in job units, where (A) is a diagram showing the left half and (B)
Is a diagram showing the right half.

FIG. 7 is a diagram illustrating an output example of effect prediction diagnosis result information for each parallel job step group.

8A and 8B are diagrams illustrating output examples of effect prediction diagnosis result information in job step units, where FIG. 8A is a diagram illustrating a left half, and FIG. 8B is a diagram illustrating a right half.

FIG. 9 is a diagram showing an output example of effect prediction diagnosis result information in data set units.

FIG. 10 is a diagram showing an output example of an effect prediction list.

FIG. 11 is a diagram illustrating a specific example of a JCL for a parallel job step.

FIG. 12 is a diagram showing data transfer between parallel job steps by a pipe data set.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 ... Parallel processing effect prediction apparatus 2 ... SMF data set 3 ... JCL library 4 ... Procedure library 5 ... Effect prediction list

Claims (11)

[Claims] 1. A method for predicting the effect of a parallel processing procedure, in which a computer performs parallel processing of a plurality of steps which can be executed in parallel, comprising: acquiring an SMF (system management function) record when the processing procedure is serially processed. Calculating a serial processing elapsed time when the processing procedure is serially processed from the SMF record; predicting and calculating a reduction time when the processing procedure is processed in parallel from the SMF record; A method for predicting the effect of a parallel processing procedure, comprising the steps of predicting and calculating the parallel processing elapsed time by the parallel processing by subtracting the reduction time, and predicting and diagnosing the effect of performing the parallel processing on the parallel processing procedure. 2. The method for predicting the effect of a parallel processing procedure according to claim 1, further comprising a step of editing and outputting a prediction diagnosis result of an effect of performing the parallel processing procedure in parallel. 3. A plurality of processing procedures each comprising a plurality of processing procedures capable of executing parallel processing consisting of a plurality of steps,
The plurality of processing procedure groups form, by a computer, a parallel processing procedure group consisting of a first parallel processing procedure group for taking over data in a pipe data set and a second parallel processing procedure group for not taking over data in a pipe data set. In the method for predicting the effect of a parallel processing procedure group for performing parallel processing, an SMF record obtained when the parallel processing procedure group is serially processed is collected, and a serial processing elapsed time when the parallel processing procedure group is serially processed is obtained from the SMF record. Calculating, based on the SMF record, determining a first elapsed time when the first parallel processing procedure group is processed in parallel by summing the respective predicted elapsed times of the processing procedures forming the first parallel processing procedure group; The second elapsed time when the second parallel processing procedure group is processed in parallel is calculated by calculating each predicted progress of each of the processing procedures forming the second parallel processing procedure group. The parallel processing elapsed time of the parallel processing procedure group is determined from the maximum value of the determined first elapsed time and the second elapsed time, and the parallel processing elapsed time is determined from the serial processing elapsed time. Predicting and calculating the effect of performing the parallel processing on the parallel processing procedure group, each of which includes a step of predicting and calculating the reduction time reduced by the parallel processing by subtracting a processing elapsed time, and effect prediction of the parallel processing procedure. Method. 4. The method for predicting an effect of a parallel processing procedure according to claim 3, further comprising a step of editing and outputting a result of predicting and diagnosing the effect of performing the parallel processing on the group of parallel processing procedures. 5. A processing time when a compound processing procedure having a plurality of said parallel processing procedures is serially processed by a computer is calculated by summing up a parallel processing elapsed time of each of the parallel processing procedures forming said composite processing procedure. The method for predicting the effect of a parallel processing procedure according to claim 3. 6. A processing procedure capable of executing parallel processing consisting of a plurality of steps comprises a first step group consisting of a plurality of steps for taking over data in a pipe data set and a plurality of steps not taking over data in a pipe data set. In the method for predicting the effect of a parallel processing procedure in which a first step group is processed in parallel by a computer, a SMF record obtained when the processing procedure is serially processed is collected from the SMF record. Effect prediction of a parallel processing procedure, comprising steps of predicting and calculating the reduction time of the first step group when the first step group is processed in parallel, and predicting and diagnosing the effect of parallel processing of the first step group. Method. 7. The method for predicting an effect of a parallel processing procedure according to claim 6, further comprising a step of editing and outputting a result of the prediction diagnosis of the effect of performing the first step group in parallel. 8. When the processing procedure includes a plurality of first step groups, the reduction time of the processing procedure is predicted and calculated by summing the respective reduction times of the first calculation step groups. Calculate the serial processing elapsed time when the processing procedure is serially processed from the record, and predict and calculate the parallel processing elapsed time when the processing procedure is processed in parallel by subtracting the reduction time from the serial processing elapsed time. 7. The method for predicting the effect of a parallel processing procedure according to claim 6, comprising a step. 9. A JC describing an instruction of each step of the processing procedure in addition to the analysis of the collected SMF record.
A step of analyzing L (job control statement) to determine a group of steps for taking over data in the pipe dataset;
The method for predicting the effect of a parallel processing procedure according to claim 1, further comprising: 10. A recording medium for causing a computer to execute a method for predicting the effect of a parallel processing procedure in which a computer performs parallel processing executable in a plurality of steps, wherein the computer executes the processing procedure serially. A SMF record at the time of processing is collected, a serial processing elapsed time when the processing procedure is serially processed is calculated from the SMF record, and a reduction time when the processing procedure is processed in parallel from the SMF record is predicted and calculated. A program for predicting and diagnosing the effect of parallel processing of the parallel processing procedure for executing each step of predicting and calculating the parallel processing elapsed time by the parallel processing by subtracting the reduced time from the serial processing elapsed time. A computer-readable recording medium characterized by the above-mentioned. 11. The recording medium according to claim 10, wherein a program for executing a step of editing and outputting a prediction diagnosis result of an effect of performing the parallel processing procedure in parallel is recorded.